Canvas Feature QEMU Code Size Report¶

Commit: 09190fa
Date: 2026-04-02T11:13:37.925806
Build target: APP=HelloSizeAnalysis APP_SUB=canvas_path_probe PORT=qemu CPU_ARCH=cortex-m0plus
Measurement method: compile a dedicated probe app, then rewrite app-local feature override + probe config headers to enable one feature set.
Scope: feature-level static qemu ELF sections only (.text/.rodata/.data/.bss).

说明¶

这份报告回答的是“整项 canvas 能力值不值得打开”，不是单个函数的边际成本。
baseline 会明确关闭 MASK / QOI / RLE 这类有全局开关的能力，再和 feature-on 变体比较。
对于没有全局总开关的能力，例如 IMAGE_TRANSFORM、TEXT_TRANSFORM、GRADIENT，这里使用代表性 probe 场景来近似描述该 feature 的整体成本。
HQ 的 line/circle/arc 路径不在本页统计范围内，请结合 hq_size_report.md 一起看。

Variant	Text	Rodata	Data	Bss	Total ROM
BASELINE	11420	568	24	5920	11988

Feature	Delta Text	Delta Rodata	Delta Data	Delta Bss	Delta ROM
QOI_CODEC	+7132	+136	+0	+28	+7268

Feature	Included Scenes	Description
QOI_CODEC	`QOI_DRAW`	QOI codec 能力，口径是打开 codec 并强制链接一个代表性绘制场景后的整体增量。

Variant	Feature Config	Text	Rodata	Data	Bss	Total ROM
BASELINE	`-DEGUI_CONFIG_FUNCTION_SUPPORT_MASK=0 -DEGUI_CONFIG_FUNCTION_IMAGE_CODEC_QOI=0 -DEGUI_CONFIG_FUNCTION_IMAGE_CODEC_RLE=0`	11420	568	24	5920	11988
QOI_CODEC	`-DEGUI_CONFIG_FUNCTION_SUPPORT_MASK=0 -DEGUI_CONFIG_FUNCTION_IMAGE_CODEC_QOI=1 -DEGUI_CONFIG_FUNCTION_IMAGE_CODEC_RLE=0 -DEGUI_SIZE_PROBE_QOI_DRAW_PATH=1`	18552	704	24	5948	19256

python scripts/size_analysis/canvas_feature_size_to_doc.py

Raw JSON is written to output/canvas_feature_size_results.json.